Natural Language Interface to Database for Data Retrieval and Processing
Though many studies related to natural language interface to a database have been conducted for many years, the results of these studies are not covered in many used cases such as the use of negative sentences, processing functions, and variety of sentence patterns with various types of query specification. To solve these problems, a model called “Natural Language Processing for Data Retrieval and Processing (NLP-DRP)” was proposed. A new algorithm named ‘Ranking Trie’ was implemented with the combination of Pattern Parsing, Ontology, and Fuzzy system to improve Lexical analysis, Semantic analysis, and Output transformation processes to allow users to retrieve and process data with various patterns of sentences and conditions. The model was incrementally tested and updated by a Learning dataset collected from users with a total of 3,868 Natural Language Query Sentences (NLQSs) then finally evaluated by the test dataset with a total of 500 NLQSs. The results showed that the NLP-DRP could retrieve data, processed, and generated outputs which consistent with user requirements with all values of Accuracy, Precision, Recall, and F-measure higher than 0.9.
- There are currently no refbacks.